Compared to version 0.4.1, the following is new:

• Fixed various errors in setup.py when pip installing
• Adjusted setuptools.setup project name to match name on PyPI

Contributors:

Anson Doan, AnHai Doan

Compared to version 0.4.0, the following is new:

• Dropped support for Python 3.6.
• Added support for Python 3.10 and 3.11.
• Switched from Nose to vanilla Unittest.
• Replaced Travis and Appveyor CI testing with Github Actions.

Contributors
Anson Doan

Compared to version 0.3.3, the following is new:

• Dropped support for Python 2 and 3.5.
• To support Python 3.8, updated the function
  py_entitymatching.matcher.matcherutils.impute_table() to use current scikit-learn's
  SimpleImputer; see issue #127.

Contributors
Matt Christie

v0.3.3 - 10/26/2020

Compared to version 0.3.2, the following is new:

• Started tracking release changes in CHANGES.txt.
• Minorly updated usage of Pandas to prevent a dependency on downgraded versions.
• Added stricter scikit-learn dependency requirements to preserve old Imputer API; will change in future release.
• Users can now provide metadata for blackbox features (ex. left_attribute, right_attribute, etc.).
• This is the last version of py_entitymatching that will support Python 2 and Python 3.5.

Contributors
Matt Christie, Diogo Munaro Vieira, Franco Venturi, Sean Soutar

v0.3.2 - 06/05/2019

Compared to version 0.3.1, the following is new:

• Added sorted neighborhood (SN) blocker.
• Re-included an older version of the debug blocker as a backup.
• Cython version was updated. The package is now built with updated Cython version ≥ 0.27.3.
• Added support for Python 3.7 and dropped support for Python 3.4.

Contributors
Pradap Konda, Phil Martinkus, Matt Christie, Chakshu Ahuja

0.3.1 - 05/10/2018
Compared to version 0.3.0, the following is new:

• Scaled up down sampler

Contributors
Pradap Konda, Phil Martinkus

v0.3.0 - 11/26/2017
Compared to version 0.2.0, the followings are new:

• Updated debug blocker code to use Cython
• Added rule-based matcher and rule-based trigger
• Updated new labeler with unit test cases

Contributors
Pradap Konda, Phil Martinkus, Pavan Kemparaju, Xin Jin, and Daniel Abraham

v0.2.0

1. Updated the down sampler command to let the user to set the random seed.
2. Updated documentation to profile dataframe using pandas-profiling package.
3. Added support for exploring pandas dataframe using pandastable and openrefine wrappers.
4. Updated the GUI code to use PyQt5.
5. Added experimental labeler with better GUI.
6. Updated auto feature generation module to validate inferred attribute types with the user.
7. Updated the select_matcher to show results for multiple metrics.
8. Added support for XGBoost classifier.
9. Fixed bugs based on the feedback from CS838 students and RIT.
10. Fixed documentation errors report by CS838 students.

v0.1.0 - 01/12/2017

• Initial release.
• Contains commands to support matching two tables using supervised learning.
• Specifically, it supports 12 different steps (such as down sampling, blocking, debugging blocker,
  matching, debugging matcher output, evaluating predictions, etc.) to come up with an entity matching
  workflow.

Contributors
Pradap Konda, Han Li, Sanjib Kumar Das, and Yash Govind